A low-cost vision-based weld-line detection and measurement technique for robotic welding

Abstract

Abstract To make robotic welding more flexible and intelligent, artificial intelligence-based systems are one of the most important developments. With the help of computer vision, an algorithm for detection of weld path, and measurement of weld gap, and weld length is developed and its implementation in robotic welding is explored in this paper. A novel method that makes use of various image processing algorithms and mathematical operations is proposed for subtraction of boundary edges and determining the weld length and weld gap at every seam point. An evaluation of the robustness of the developed algorithm was performed on butt joints of various shapes (e.g., straight, zigzag, and curve) and sizes (e.g., length of the joint, gaps between abutting edges). The obtained length and gap values are validated by physical measurements, and the absolute errors were found to be within ± 1 mm, and ± 0.1 mm, for length and gap measurements, respectively, The model possesses 99.5% average precision in detecting atypical weld region. Weld seam information can be extracted effectively with the proposed algorithm, which proves its viability for industrial applications.